We promised to publish deliverability stats once a quarter, and we're keeping that promise — including the uncomfortable parts. Q2 closed with an average success rate of 96.4%, up 1.1 points from Q1, but with a notable dip in May.

By region, the picture is about what you'd expect. Europe holds at 97.8% with a median time-to-code of nine seconds — stable pools, predictable carriers. The Americas came in at 96.9% at eleven seconds. Asia delivered 94.2%: the spread between countries within the region is wider than the spread between regions themselves. Africa sits at 95.1%, with the fastest quarter-over-quarter growth of any region.

May dragged down the Southeast Asia average. Two large pools in Indonesia and the Philippines started losing messages from messaging apps — success there dropped as low as 88%. The cause turned out not to be on our side (the services changed their filtering), but that's cold comfort to the user, so we rewrote the selection logic.

Pool priority used to be calculated from a full day of statistics. That window is too sluggish: degradation only becomes visible by evening, and until then orders keep flowing into the failing pool. Now the window is short and rolling, and recent failures carry more weight. A pool that starts losing messages drops in priority within tens of minutes instead of a full day.

The second change: automatic refunds got faster. We shortened the waiting timeout for pools with falling stats — if the system sees the odds of a code are low, there's no reason to hold the user's money for twenty minutes. By the end of June, the region's success rate was back up to 95.6%.

In Q3 we plan to surface deliverability per service right in the catalog, not just in the report. A number refreshed once an hour will help you pick a country before you buy, instead of after a run of failed orders.